#Introduction When I was younger, I used to play a game called Runescape. It had a centralized trading platform called the Grand Exchange. Players (there are over 200 million registered accounts now according to Wikipedia) would submit buy and sell orders for most tradeable items in the game; back then, and like most people now, I produced goods and sold them for money through this platform. It's also obvious though, that the Grand Exchange functions somewhat like a stock market. There are several differences, but many claim that players can make a profit from trading on the Grand Exchange. In this project, I plan to obtain data for various items traded in the Grand Exchange and analyze the data. The goal is to look for trends and relationships over time and between related items in an attempt to better predict how prices will move; this will be a good opportunity to experiment with machine learning algorithms and gain a stronger understanding of the overall process from gathering data to final results.
Please read Grand-Exchange-Analysis-Notes.pdf for more information.
#Process There will be several steps to take in this project, and as I am a bit new to machine learning, I do not have an exact process outlined yet. I do know that I will need to
- Write code to download data about specific commodities from the Runescape Grand Exchange database.
- Determine what commodities are promising to analyze. For example, common, useless items have no volatility or volume on the Grand Exchange and there's no profit to be made from trading them.
- Of the commodities that are promising to analyze, generate hypotheses as to what relationships between certain commodities might exist. Perhaps prices might also follow patterns with the time.
- Run machine learning algorithms and test out those hypotheses.
- Determine how to quantify the "success" of some output of the algorithms.
- State any findings.
- Go back to step 3 if the process was enjoyable :D .
As of right now, I have procured price data for about 2000 items and am experimenting with algorithms to determine which ones are promising to trade with.